Global optimization method using SLE and adaptive RBF based on fuzzy clustering

摘要: High fidelity analysis models, which are beneficial to improving the design quality, have been more and widely utilized in modern engineering optimization problems. However, high models so computationally expensive that time required is usually unacceptable. In order improve efficiency of involving can be upgraded through applying surrogates approximate greately reduce computation time. An efficient heuristic global method using adaptive radial basis function (RBF) based on fuzzy clustering (ARFC) proposed. this method, a novel algorithm maximin Latin hypercube successive local enumeration (SLE) employed obtain sample points with good performance both space-filling projective uniformity properties, does great deal metamodels accuracy. RBF adopted for constructing metamodels, increasing number approximation accuracy gradually enhanced. The c-means applied identify reduced attractive regions original space. numerical benchmark examples used validating ARFC. results demonstrates most application optima effectively obtained comparison response surface (ARSM) proves proposed intuitively capture promising efficiently or near-global optimum. This improves convergence problems, gives new strategy problems models.